funasr/models/e2e_asr_paraformer.py
@@ -370,19 +370,10 @@ encoder_out, encoder_out_lens ) assert encoder_out.size(0) == speech.size(0), ( encoder_out.size(), speech.size(0), ) assert encoder_out.size(1) <= encoder_out_lens.max(), ( encoder_out.size(), encoder_out_lens.max(), ) if intermediate_outs is not None: return (encoder_out, intermediate_outs), encoder_out_lens return encoder_out, encoder_out_lens return encoder_out, torch.tensor([encoder_out.size(1)]) def calc_predictor(self, encoder_out, encoder_out_lens):
